- Google’s AI pivot is quietly killing the click economy it spent 30 years building
- Companies that don’t rethink how AI finds them risk disappearing from search entirely
Alvin Koay (pic) has a question he likes to ask CMOs when they come to him wondering why their search traffic has collapsed. Not what keywords are you targeting, or how many backlinks do you have. He asks: when you type a query into Google today, do you even see the results page?
Most pause. Because they know the answer. What greets them now is an AI-generated summary sitting at the very top of the page, a paragraph of synthesised information that answers the question before a single link is clicked. On mobile, it fills the entire screen. Traditional search results, if visible at all, are somewhere below the fold.
“The real estate on the search results page has been taken over by answers,” said Alvin, founder and CEO of GrowthPro, a Malaysian digital marketing agency he launched in 2018, that pivoted to specialise in Answer Engine Optimisation, or AEO in mid 2024. “And when that happens, nobody clicks through to your website. The traffic just stops.”
This is not a niche complaint from publishers. It is a structural shift playing out across every industry that built its digital strategy around search engine rankings, and one that Alvin contends most SEO agencies are still unwilling to honestly explain to their clients.
The numbers behind the silence
Global publisher traffic from Google search dropped by a third in the year to November 2025, according to tracking by Press Gazette and Chartbeat. In the US, organic referrals from Google were down 38% year-on-year. CNN lost between 27% and 38% of its traffic. Business Insider shed 55%. Forbes and HuffPost each lost roughly half their Google-driven audience.
The culprit is Google’s AI Overviews, the AI-generated summaries that now appear above organic results (use pix to show this) for a growing share of queries. When an AI Overview is present, organic click-through rates drop 61%, from 1.76% to 0.61%, according to Seer Interactive’s September 2025 study tracking 25.1 million impressions across 42 organisations.
Paid click-throughs fell 68% for the same queries. Google’s AI Mode, the full answer engine the company launched in mid-2025 which is a separate, more conversational AI search experience, takes this further: around 93% of searches conducted in AI Mode end without a click to an external website, or what is called zero-click rate. On mobile, it sits at 77%.
The most disorienting part for marketing teams is that rankings and impressions can still look healthy even as actual traffic falls. BrightEdge found that search impressions rose 49% year-on-year in May 2025, but average click-through rates dropped 30% over the same period.
In other words, publishers and brands may still be visible on Google, but users are increasingly getting what they need from the AI answer itself instead of clicking through to the original website.
“For many months, CMOs have been watching their traffic tank and spending more money trying to fix it,” Alvin said. “Most SEO agencies cannot explain what is happening. The answer is simple: Google is no longer a traffic machine.”
This may be hard to understand as its entire monetisation logic, built around keyword-matched intent, depended on users clicking through to external pages. But now, every improvement to AI summaries short-circuits that revenue loop.
“Google is a search engine but it needs to become an answer engine because search behavior has changed. Now instead of sending people to websites, they keep people in the walled garden which is the AI Overview or AI Mode (answer engine),” said Alvin.
Google’s situation was not entirely of its own choosing. OpenAI’s launch of ChatGPT in Nov 2023 caught the company flat-footed despite having DeepMind and decades of AI research behind it. Gemini and AI Mode are the response. But the cost of staying relevant in the answer engine era is a direct cannibalization of the search ad model that has funded everything else.
Be that as it may, the expected cannibalization some expect has not happened yet as evidenced by Google’s revenue (see table) in the four quarters since May 2025 (with March 2025 taken as the baseline because it was in May 2025 that AI Mode was launched globally) where ‘Google Search & other’ as a category rose from US$50.7 billion to US$60.4 billion, up about 19% year-on-year.
Still, the shift in user behaviour is measurable. Gartner predicted in early 2024 that traditional search engine volume would drop 25% by 2026 due to AI chatbots and virtual agents. Most analysts now consider that projection conservative.
Why ranking on Google’s first page no longer guarantees AI visibility
According to Alvin, understanding what has changed under the hood is critical because optimising for AI answer engines is fundamentally different from optimising for traditional Google Search.
“For years, search engines worked as ranking algorithms,” he explains. “They evaluated millions of pages against hundreds of signals and produced an ordered list of links. SEO was about scoring highly enough to appear near the top so users would click through to your website.”
Large language models (LLMs) such as ChatGPT, Gemini and Google’s AI Mode operate differently. They function primarily as retrieval and synthesis systems rather than ranking engines. Instead of presenting users with a list of pages to browse, they retrieve relevant pieces of information from multiple sources, combine them with their existing knowledge, and generate a direct answer.
For a single conversational query, an LLM may break the request into dozens of smaller searches across multiple entities before assembling a response.
If the content it finds is a long article written mainly for human readers, without clear structure, headings or machine-readable organisation, the model may struggle to identify the relevant information and simply move on to another source.
“A lot of legacy SEO content is now a liability for answer engines because it is not structured properly,” Alvin said. “Content that ranks on page one in Google often does not show up in ChatGPT at all. Content from page five or six might be exactly what the answer engine cites because it is chunked correctly, uses the right headers, FAQ structures and adopts entity-based schema.”
The data reflects that shift. BrightEdge reported a 400% increase in citations pulled from results ranked between positions 21 and 30, while 89% of AI citations now come from pages outside Google’s top 100 organic listings. The overlap between Google’s top-10 rankings and AI Overview citations has fallen sharply, from 75% in mid-2025 to between 17% and 38% by early 2026.
In short, ranking highly on Google no longer guarantees that AI systems will discover, trust or cite your content.
New rules require new metrics beyond rankings and clicks
If rankings, impressions, clicks and dwell time are becoming less reliable indicators of online visibility, what should businesses measure instead?
Alvin says the focus must shift from whether users click on a webpage to whether AI systems mention, trust and recommend a brand when answering questions.
GrowthPro now tracks four key measurements for clients.
The first is Share of Voice — how often a brand appears in responses generated by ChatGPT, Gemini, Perplexity, Google AI Overview and Grok compared with competitors.
The second is citation volume, which measures how frequently answer engines quote or attribute information from the company’s own website.
The third is citation positioning — whether the brand is mentioned first, fifth or not at all when AI generates an answer.
The fourth is sentiment, which measures how AI platforms describe the brand based on information gathered from across the web.
“LLMs ingest data from every digital touch point,” Alvin said, trying to explain the sum of the four measurements. “Your website, your social media, your Lazada store, your press mentions. If your website says one thing and your social media says something different, the LLM cannot trust you. And if it cannot trust you, it either hallucinates or it skips your content entirely.” That’s a new reality sure to send shivers of fear down the spines of marketing leaders.
That means brand consistency is no longer simply good marketing practice. It has become a technical requirement if companies want AI systems to confidently retrieve and recommend their content.
Some CMOs are already catching on to this new reality, noting that AI search places much greater weight on user-generated discussions about brands than traditional search ever did.
Freshness also matters. New York-based Generative AI marketing intelligence firm, Profound, found that content published or updated within the previous 13 weeks is 50% more likely to appear in chatbot responses. Well-structured, recently updated content can therefore leapfrog competitors whose websites still rely on years-old SEO articles written for Google’s earlier ranking system.
As Alvin puts it: “Companies who have been losing in SEO have a real opportunity right now. If they deploy AEO strategies properly, they can move past competitors fast because the old content that gave incumbents an advantage is now working against them.”
The gold rush, and what to watch for
Not surprisingly, the urgency around AEO has produced a market. The global SEO and AEO services market are projected to reach US$171 billion by 2030, up from US$81.4 billion last year, according to MarkNtel Advisors. Adobe’s acquisition of Semrush for US$1.9 billion in December 2025 was framed in part around AEO measurement capabilities, with Adobe president Anil Chakravarthy noting that every CMO today is thinking about how their brand shows up in ChatGPT.
Yet, in Malaysia, Alvin is sceptical of the market impact. His concern is not with AEO as a discipline but with agencies claiming to offer it while running the same practices that were already failing.
“Nine out of ten agencies are still doing link building, which is against Google’s own terms of service and has been for years,” he said. “Now they have slapped AEO onto the same pitch. They optimise keywords, buy backlinks, and call it answer engine optimisation. They do not even have a tool to track AI visibility.”
To address this very gap in the market, GrowthPro built Zicy, an AI-powered platform that tracks share of voice, citation volume, positioning and sentiment across five major answer engines, and uses an AI agent to identify content gaps and restructure website content for LLM retrieval.
For companies evaluating AEO vendors, Alvin offers three questions worth asking: What tools do you use to track AI visibility across answer engines? What specific KPIs will you report on, and do those include Share of Voice, citations and sentiment, or just rankings and traffic? And how do you structure content for LLM retrieval, not just human readability? “If the answers revolve around keywords and backlinks, walk away,” he says.
Not everyone is taking such a clear-cut approach. Some agency leaders believe that AEO and SEO will likely merge rather than one replacing the other, as both kinds of search persist. The early predictions that AEO would eliminate SEO has changed into a more pragmatic view.
AI platforms continue to grow rapidly. ChatGPT reached about 1.19 billion monthly active users by early 2026, while the broader generative AI ecosystem exceeded 1.57 billion users outside mainland China. Gemini’s referrals to publisher sites also rose 388% year on year between September and November 2025.
Yet, as Alvin said, outbound traffic is moving in the opposite direction. Similarweb found AI platforms referred 15% fewer US visitors to external websites in January 2026 than three months earlier, while referral rates fell from 18.8% to 15.8%. More users are relying on AI-generated answers without clicking through, reinforcing his view that these platforms are becoming answer engines rather than traffic routers.
For Malaysian businesses, Alvin’s advice is straightforward. Ask ChatGPT, Perplexity and Google AI Mode the questions you think your customers would ask. And, if you find that your brand isn’t being recommended, it is time to hit code red and audit every digital touchpoint, from your website and social media to marketplace listings and press coverage, to ensure they present
“The winners will be brands whose content is structured, consistent and authoritative enough for AI to retrieve and recommend,” stresses Alvin.
AI was used to aid in the writing of the first draft with the writer and an editor responsible for the final version.
Related Articles
Keyword(s) :